variable rate steering racks absolutely exist - early 2000s BMW 5 series and honda S2000 are the ones off the top of my head.
They are not common. Mostly nowadays a similar effect is created via variable assist. Human interface-wise, nonlinear controls are REALLY not a great idea without specific training.
If you look at airplanes, where pilots are type rated, you have a lot more controls mixing with electronic flight controls and various regimes. The yoke/stick might control pitch at one point, angle of attack at another, altitude in another, and control surface deflection in another...or more realistically a mix of all of the above. Multiply by 6 dimensions of control.
I think about my mother. Recently we rented a Subaru Outback. I was comfortable with the minor force/lane centering control actions it took while on cruise control. She on the other hand couldn't handle it - it became a negative feedback loop. It woudl correct her into the cetner of the lane, and she responded to it as 'the car pulling' and would over react and put us in the next lane. The car correcting itself was disconcerting and provoked a corresponding reaction that was programmed in her brain as 'correct' because the heuristic was 'force through the wheel = something wrong'.
The issue here isn't whether the car does something good or bad, it is whether it does something that the average driver reacts to in a safe an appropriate way. That comes from meeting their preconceptions as to how steering behaves, unless Tesla wants to argue for type-certing their owners.
They are not common. Mostly nowadays a similar effect is created via variable assist. Human interface-wise, nonlinear controls are REALLY not a great idea without specific training.
If you look at airplanes, where pilots are type rated, you have a lot more controls mixing with electronic flight controls and various regimes. The yoke/stick might control pitch at one point, angle of attack at another, altitude in another, and control surface deflection in another...or more realistically a mix of all of the above. Multiply by 6 dimensions of control.
I think about my mother. Recently we rented a Subaru Outback. I was comfortable with the minor force/lane centering control actions it took while on cruise control. She on the other hand couldn't handle it - it became a negative feedback loop. It woudl correct her into the cetner of the lane, and she responded to it as 'the car pulling' and would over react and put us in the next lane. The car correcting itself was disconcerting and provoked a corresponding reaction that was programmed in her brain as 'correct' because the heuristic was 'force through the wheel = something wrong'.
The issue here isn't whether the car does something good or bad, it is whether it does something that the average driver reacts to in a safe an appropriate way. That comes from meeting their preconceptions as to how steering behaves, unless Tesla wants to argue for type-certing their owners.